Results
The proportion of individuals which survived till the end of the reproduction period in the GTI treatment declined with the increasing temperature (Fig. 1), and the decline was significant when compared to control treatment (Treatment * Generation est. ± s.e. = -1.004 + 0.228, z = -4.393, p < 0.001; Table S1). Within GTI populations, the survival was affected by the interaction between population type (FT vs. ST) and generation (est. ± s.e. = 0.538 + 0.184, z = 2.992, p = 0.003; Fig. 1, Table 1). Strikingly, among populations subjected to temperature stress, very few adults survived in FT populations at generation four (Fig. 1) and consequently, nine (75%) of these populations went extinct (Fig. 2). At the same time, only one (8,3%) ST population went extinct, a significantly lower proportion compared to FT populations (est. ± s.e. = 3.724 + 1.741, z = 2.139, p = 0.032). Interestingly, extinction proceeded despite that expression of fighter morphs in FT populations declined from 76% at first generation to 25% at fourth generation (est. ± s.e. = -1.598 + 0.218, z = -7.327, p < 0.001; Fig. S2, Table S2). Indeed, the reason for higher extinction of FT did not seem to have been associated with higher male mortality in FT populations, as indicated by the non-significant effect of population type*sex interaction on proportion of surviving adults across four generation (full model: est. ± s.e. = 0.154 + 0.301, z = 0.511, p = 0.609). However, there was a significant generation * sex interaction (best model: est. ± s.e. = 0.462 + 0.145, z = 3.184, p < 0.001 Fig. S3; Table S3): in both F and S populations, females survived better than males in early generations, but then in following generations, as temperature increased and survival decreased, the differences between sexes decreased.